In this article, Mohammad Golshani addresses the Quantitative Risk Assessment of hydrogen sulphide hazards.
Personal Protective Equipment (PPE) such as self contained breathing apparatus, protective clothing, safety glasses, boots and gloves are widely used in today’s oil and gas industry. Temporary refuge shelters as well as fresh air supply systems are normally recommended when the presence of highly toxic gases is possible.
The question is how to avoid a large number of fatalities or serious injuries in the case of undesired toxic gas release or toxic liquid spill, and how deeply should one look into the safety equipment and systems? With companies in the oil and gas industry operating globally they are always worried about their reputations, as huge markets could be lost. Companies are therefore concerned about the health of employees, the impact of their business on the environment, the capital investment from one side and the amount of money they will need to invest in order to practically and reasonably avoid a catastrophic release of toxic substances.
Some may believe the greater the investment in safety, the safer the industry will be. While this may be common sense, history shows this isn’t true. In this article we will use a risk based methodology to try to find an optimised solution, which takes into consideration the reasonable amount of investment required for safety in the industry, to avoid the occurrence of a Major Accident Hazard (MAH). We will do this by identifying components that contribute to these hazards, as well as suggesting methods to either eliminate these components, or provide recommendations that make the risks tolerable.
The risk based approach is fully quantitative, meaning that the final figures from the risk assessment can be compared against the criteria established for this purpose. It consists of a robust methodology, including mathematical modelling of mass transfer phenomena occurring during a gas release or liquid spill, as well as a comprehensive frequency analysis that shows the probable frequency of a MAH occurring.
An essential part of this analysis is the range of tolerable and intolerable risk figures. Many stack holders contribute to risk criteria, but some key role players such as large operators and governmental sectorsmay have large databases for accident histories. These histories include the failure frequency of equipment and other components in the oil and gas processes that may establish their own risk criteria. This method, known as Quantitative Risk Assessment (QRA), is now widely used to quantify risks.
QRA is a quantitative approach focused on assessing fatality risks, considering all possible events such as fire, thermal radiation, explosion and toxic substance release in gas or liquid form, thereby giving an overall figure of risk levels that the workers or public are exposed to during normal plant operation.
QRA studies are divided into:
• Consequence analysis
• Frequency analysis
At the end of the study the findings from these two sections are integrated, resulting in a final estimation of risk figures that can then be compared with the risk criteria. Risk reduction measures can then be recommended, with the final risk figures updated after recommendations are implemented.
Let’s consider a toxic material such as hydrogen sulphide (H2S). When present in plant feed streams it can cause very dramatic and irreversible health effects, even in concentrations as low as 100 parts per million (ppm).
The following sections explain each step of the QRA for H2S.
Consequence analysis involves targeted modelling of both the physical explanation and mathematical formulation of a MAH, such as the release of H2S from a vessel or pipe.
Many physical observations and mathematical correlations are developed by scientists and engineers to understand and modeldifferent steps in MAH occurrence. These models are then turned into mathematical correlations in order to correctly programme the computers. The meticulously designed computer software programmes available on the market today can easily estimate the extent and severity of the impact of a MAH on workers, the public, the environment and a plant’s capital investment.
To calculate the extent and severity of a potential H2S gas release, all plant systems must be sectioned based on the location of the isolation valves, or emergency shutdown valves (ESDV). These are normally used to segregate material inventories in different parts of the plant during catastrophic events, so that additional materials cannot be consumed by the incident in a fire, for example, exacerbating the situation.
Sectioning also helps to estimate how much H2S will be released before the rate decreases to minimum due to pressure equalisation between the leaking streams and atmosphere, as well as how long it would take for it to stop. Since guessing the size of the leak’s source is possible, hole size scenarios of 5mm, 25mm and 100mm up to full bore rupture are modelled, to predict the shape, size and orientation of potential hazard zones.
Meteorological data plays a very big role in the dispersion of H2S in the environment. If it is windy then large pockets of H2S will be moved to the surrounding areas, much further and quicker than when the weather is stable. Warmer temperatures will cause more movement in the layers of the air, again causing greater dispersion of the toxic gas. Humidity, on the other hand, may have a different effect. Higher humidity could limit the spread of H2S into the layers of air adjacent to the source, while lower humidity would not offer the same resistance.
Just as a variety of leak sizes must be accounted for, a range of weather stability classes must be considered to cover the uncertainty of weather conditions in the event of H2S release. Usually, a combination of Pasquill Stability Classes and wind speeds are used for this purpose, including 2F, 3B, 5D and 15D. Stability classes start from A, which indicates extremely unstable conditions, B presents unstable conditions, F to show moderately stable conditions and D to signify neutral scenarios.
The effects of a catastrophic event such as H2S release will depend on which direction the toxic cloud may travel. To compensate the uncertainty of the cloud movement direction, wind rose information is programmed into the software to predict the most probable risk of exposure scenarios for the hazard zones affecting the workers and public. The probability of fatality due to different concentrations of H2S is also calculated. This is usually estimated using probit functions, which are based on laboratory and experimental results from tests conducted on H2S effects in different concentrations.
Frequency analysis is mainly used to estimate the number of times an undesired accident can happen in a year. Using the results of this analysis, it is possible to estimate risk figures of MAH and gain knowledge of the severity and probability of an incident.
Each component within an isolatable section has its own unique frequency of failure, which in this case is the frequency of leakage. These figures are based on the history of failure in similar components during the years of operation. Usually these figures are kept in databases and published by organisations such as the UK’s Health and Safety Executive (HSE) or the American Institute of Chemical Engineers (AICHE).
To estimate the frequency of a leak in an isolatable section, a piping and instrumentation diagram must be used alongside piping isometrics to identify the exact length of the piping. The number of components and pieces of equipment present must also be known.
The frequencies of all components in one section will then be extracted from the database and added together to calculate the total frequency of the leak in a specific area. The whole process, from identifying the parts in one section to deriving the data and estimating the total frequency of the leak, is called the parts count.
Since we are looking into H2S gas, other parameters contributing to the risk of occurrence of MAH such as fire, explosion and flash fires, which are basically the probability of immediate and delayed ignition, are not taken into account. Leak frequency appointment for the initial sizes of 5mm, 25mm, 100mm and full bore rupture will therefore be only a summation of the failure rates of all the items in an isolatable section for that specific hole size.
Sometimes the probability of successful or unsuccessful isolation will be considered; for example, if the automatic isolation process is successful it will take approximately two minutes to close the ESDV. This additional two minute flow of material from upstream will therefore be factored into the isolatable section. If automatic isolation is unsuccessful, this duration will change to ten minutes, as this is the time required for manual isolation.
It is important to note that all the previously mentioned calculations will be done with software programmes, since the calculations are too many and too complex to conduct manually. Correctly inputting the basic information is at the heart of this process and this requires a lot of hard work and focus.
Plant risks are defined using numerous terms and acronyms. These can be divided primarily into individual and societal risks. Individual risk
Location specific individual risk (LSIR) shows the risk of a particular location if an individual were to stand in that location 24 hours a day.
Individual risk per annum (IRPA) is a combined risk to a person considering exposure to all kinds of MAH. It is calculated as the frequency of fatality per year when considering the presence of the person in a different location. The exact schedule of personnel movement and presence in the plant is required to conduct this calculation.
Potential loss of life (PLL) is the average annual number of fatalities expected among personnel. This arises from their work on the project facilities and their travel to and from project sites.
Frequency number (F-N) curves graphically depict the number of fatalities as a function of estimated frequency that those fatalities will occur. This approach allows differentiation between high impact/low frequency events and low impact/high frequency events when PLL is possible.
LSIR is usually generated to show the risk of a specific location for one undesired accident, such as H2S release, of a cumulative risk figure for all the undesired MAH that could affect a specific location.
Figure 1 shows a risk contour plot for a gas treatment facility. The contours are presented as different coloured regions on the plot, representing bands of LSIR as follows:
As shown in Table 1, the basic IRPA value is derived from the LSIR value at this location, multiplied by the base occupancy level. The base occupancy level is derived from the time spent on the site, which is based on a 12 hour shift and rotation. The revised IRPA is calculated by taking into account the percentage of time during a shift that operators spend out on the process plant. It is assumed that when personnel are not out on the plant that they are protected from most hazards by being inside a building.
Similarly, for PLL the most exposed group should be assumed to be representative of all worker groups.
As shown in Figure 2, preliminary estimates of the personnel required to run the facility should be applied.
Societal risk should not be confused with the risk to society or the risk as perceived by society. The word societal is merely used to indicate a group of people. Societal risk refers to the frequency of multiple fatality incidents, which include workers and the public. Societal risk is usually represented by an F-N curve and compared with risk criteria defined by the company, based on international references.
Finally, all the risks calculated should be as low as reasonably practicable (ALARP).
To be ALARP the residual risks should be in an acceptable region, as defined by the owner of the plant. It must also be possible to demonstrate that the cost involved in reducing the risk further would be grossly disproportionate to the benefit gained.
The principle of reducing risks to ALARP arises from the fact that infinite time, effort and money could be spent on reducing a risk to zero. It should not be understood as simply a quantitative measure of benefit versus detriment. It is instead better practise to judge and balance the risks and societal benefits.
As shown in Table 2, the maximum individual risk criteria is mostly a matter of policy, yet it is defined based on well established and robust international practises.
As shown in Figure 5, the final calculated risk figures must fall below either the acceptable area where the risk figures are less than or equal to the minimum acceptable criterion, or within the ALARP area, for which the risk figures are between the maximum and minimum acceptable range.
If the risk figures are above the maximum acceptable level, then remedial actions should be provided. In the case of H2S release, as mentioned earlier, these additional measures include PPE such as self contained breathing apparatus, protective clothing, safety glasses, boots and gloves, as well as temporary refuge shelters and fresh air supply systems. The choice of which measures to use depends on the proximity of the calculated risk to the maximum acceptable risk figure.
Published: 17th Jun 2014 in Health and Safety Middle East